osint-investigation
Pass
Audited by Gen Agent Trust Hub on Oct 10, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONNO_CODE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data (media, documents, web pages) which is a potential surface for indirect prompt injection.
- Ingestion points: Processes user-supplied artifacts like photos, video clips, documents, and public profiles as described in
SKILL.md. - Boundary markers: No specific boundary markers or delimiters are defined for interpolating these artifacts into prompts.
- Capability inventory: Utilizes the host's search, browser, file, and analysis tools as noted in
SKILL.md. - Sanitization: Includes explicit instructions to treat metadata and code snippets as evidence only and never as instructions to execute.
- [NO_CODE]: The skill contains no executable code, scripts, or binary files. It consists entirely of markdown instructions and a license text.
- [SAFE]: The skill includes strong safety guidelines prohibiting the identification of private individuals, assembly of dossiers, or bypassing of access controls.
Audit Metadata